Least-Ambiguous Multi-Label Classifier

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hagos, Misgina Tsighe, Lundström, Claes
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914035746734080
author Hagos, Misgina Tsighe
Lundström, Claes
author_facet Hagos, Misgina Tsighe
Lundström, Claes
contents Multi-label learning often requires identifying all relevant labels for training instances, but collecting full label annotations is costly and labor-intensive. In many datasets, only a single positive label is annotated per training instance, despite the presence of multiple relevant labels. This setting, known as single-positive multi-label learning (SPMLL), presents a significant challenge due to its extreme form of partial supervision. We propose a model-agnostic approach to SPMLL that draws on conformal prediction to produce calibrated set-valued outputs, enabling reliable multi-label predictions at test time. Our method bridges the supervision gap between single-label training and multi-label evaluation without relying on label distribution assumptions. We evaluate our approach on 12 benchmark datasets, demonstrating consistent improvements over existing baselines and practical applicability.
format Preprint
id arxiv_https___arxiv_org_abs_2509_10689
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Least-Ambiguous Multi-Label Classifier
Hagos, Misgina Tsighe
Lundström, Claes
Machine Learning
Multi-label learning often requires identifying all relevant labels for training instances, but collecting full label annotations is costly and labor-intensive. In many datasets, only a single positive label is annotated per training instance, despite the presence of multiple relevant labels. This setting, known as single-positive multi-label learning (SPMLL), presents a significant challenge due to its extreme form of partial supervision. We propose a model-agnostic approach to SPMLL that draws on conformal prediction to produce calibrated set-valued outputs, enabling reliable multi-label predictions at test time. Our method bridges the supervision gap between single-label training and multi-label evaluation without relying on label distribution assumptions. We evaluate our approach on 12 benchmark datasets, demonstrating consistent improvements over existing baselines and practical applicability.
title Least-Ambiguous Multi-Label Classifier
topic Machine Learning
url https://arxiv.org/abs/2509.10689